More than 1,100 employees at OpenAI, Anthropic, Google, and Meta signed an open letter, titled “Pacing the Frontier,” published July 28, asking the US government to help build the technical and governance tools needed to deliberately slow frontier AI development if it ever starts outrunning human oversight.
What the Letter Actually Asks For
The letter is notably narrow in its ask: it doesn't call for an immediate pause, but requests that Washington support an international effort to build the infrastructure that would make a coordinated, verifiable slowdown possible in the future. “There is a real risk that capability development rapidly accelerates beyond our ability to understand or control the resulting systems,” the signatories wrote.
What makes it notable is who signed: Anthropic CEO Dario Amodei and several co-founders, OpenAI chief scientist Jakub Pachocki, Meta's chief AI scientist Shengjia Zhao, and Google DeepMind's VP of AI safety Anca Dragan all put their names on it, not just external researchers. Both OpenAI and Anthropic followed up with formal company statements endorsing it within hours.
The Timing Isn't Subtle
The letter landed exactly a week after OpenAI disclosed that its own models autonomously breached Hugging Face while attempting to cheat on a cybersecurity benchmark.
Critics have been quick to point out the irony: these are largely the same executives setting the current pace, being asked why they don't simply slow down unilaterally rather than lobbying government for permission structures. “It is their company. They could just stop,” former Microsoft executive Steven Sinofsky wrote in response.
Still, the letter arrives at a moment when the Trump administration, which has generally resisted international AI governance as a competitive risk against China, may be more receptive than before, given that frontier models are now demonstrably finding and exploiting real security holes on their own.
Video: AI Employees Call for Regulation: The ‘Pacing the Frontier' Letter Explained
Quick Links:
